Research On Decoding Algorithms And Error-control Performance Of Spinal Codes

With the rapid development of wireless communication technology,the efficient and reliable data transmission technology has caused extensive concern.In the scenarios where the channel state changes rapidly or is unknown,it is difficult for the transmitter to accurately predict the channel state,so that determining the appropriate rate for information transmission is hard.Rateless codes could provide a good solution,which can continuously generate the coded symbols.When the receiver gets enough symbols,the transmitted information can be correctly recovered.There is no need to send feedback to the transmitter during the decoding process.Therefore,the rateless codes have become a research hotspot in wireless communication technology.Spinal codes,a new type of rateless codes,have a simple coding structure,strong resistance to noise and bit error.Moreover,it has been proved that spinal codes can approach Shannon limit over both binary symmetric channel(BSC)and additive white Gaussian noise(AWGN)channel.The main work of this thesis is on the study of the decoding algorithm and error-control capability of spinal codes,including the following contributions:Firstly,the thesis focuses on the decoding algorithm of spinal codes.In order to decrease the complexity,a layered unquantized Fano sequential decoding algorithm is proposed in the thesis.The algorithm improves the path metric to increase the accuracy of selecting path.At the same time,the decoding tree is divided into several layers and the unquantized Fano sequential decoding algorithm is applied to each layer to search for the optimal path.In the algorithm,the jump of the decoder occurs in the current layer instead of the entire tree,which effectively avoids the decoder to keep retracting and reduces the decoding complexity.Moreover,the memory only needs to retain the extended nodes of the current layer,which greatly reduces the use of storage space.The simulation results show that compared with the existing bubble decoding,the forward stack decoding algorithm and the modified unquantized Fano sequential decoding algorithm,the proposed algorithm has no loss of rate performance and further reduces the decoding complexity.Especially in the case of low signal to noise ratio,the complexity is relatively reduced.Secondly,the error-control performance of spinal codes is analyzed in detail.Aiming at high error probability of spinal codes,which is caused by the weak error-control capability of the last message segment,the spinal codes with the first and last echo are proposed.In the scheme,a copy of the last message segment is placed at the forefront of the message sequence and then they are sequentially encoded,so that more coded symbols transmit the last segment.Thus,the error rate of the last segment is greatly decreased and the errorcontrol capability of spinal codes is improved as a whole.The feasibility of the spinal codes is verified theoretically.Simulation results show that the error rate of the spinal codes with the first and last echo is much lower than that of the original spinal codes,and the rate performance is also improved a lot.Finally,aiming at the weak error-control capability of the original spinal codes,an improved design of spinal code is proposed,which is named as the spinal codes with unequal-blocking structure.Compared with the original spinal codes,the spinal codes with the first and last echo improve the error-control performance.However,they add a piece of coding calculation and send one more coded symbol.The spinal codes with the unequal-blocking structure can increase the rate and decrease of error rate without additional calculations.Then,the block model,coding structure and decoding principle of the unequal-blocking structure are given.Based on the block model,three simple and efficient block schemes are designed without increasing the complexity of coding and decoding.Simulation results show that the spinal codes with unequal-blocking structure using the bubble decoding improve the rate and throughput performance and reduce the error rate compared with the performance of the original spinal codes.